Recently, solar-induced chlorophyll fluorescence (SIF) is a promising tool to estimate gross primary production (GPP). Photosynthesis gradually saturates with the increasing light, but fluorescence tends to keep increasing, leading to a nonlinear SIF-GPP relationship. This nonlinearity occurs for sunlit leaves but not for shaded leaves for which photosynthesis is light-limited. However, the separation of sunlit and shaded SIF has not been systematically investigated when estimating GPP from SIF. Therefore, it is promising to develop a model for GPP estimation considering such differences. This study proposed an approach to separate the total canopy SIF emission (SIFtotal) from TROPOspheric Monitoring Instrument (TROPOMI) SIF into their sunlit and shaded components (SIFsun and SIFshade). The nonlinearity and linearity in SIF-GPP relationships for sunlit and shaded leaves were incorporated into a two-leaf hybrid model, which was fitted using flux tower data and then evaluated using leave-one-site-out crossing validation. We also elucidated the distinct SIF-GPP relationships between sunlit and shaded leaves using the Soil-Canopy-Observation of Photosynthesis and the Energy balance (SCOPE) model simulation. Compared to previously used linear (R2 = 0.68, RMSE = 2.13 gC⋅m^-2*d^-1) or hyperbolic (R2 = 0.72, RMSE = 2.01 gC⋅m^-2⋅d^-1) model based on the big-leaf assumption, our proposed two-leaf hybrid model has the best performance on GPP estimation (R2 = 0.77, RMSE = 1.79 gC⋅m^-2⋅d^-1). We also applied this two-leaf hybrid model to estimate the global GPP during the main growing season in Northern Hemisphere, which were highly correlated with several existing GPP products, with R2 ranging from 0.79 to 0.88. These results will improve our understanding of the relationship between SIF and GPP for sunlit and shaded leaves and will advance application of satellite SIF data to GPP estimation.
more »
« less
This content will become publicly available on June 1, 2025
Retrieval of Crop Canopy Chlorophyll: Machine Learning vs. Radiative Transfer Model
In recent years, the utilization of machine learning algorithms and advancements in unmanned aerial vehicle (UAV) technology have caused significant shifts in remote sensing practices. In particular, the integration of machine learning with physical models and their application in UAV–satellite data fusion have emerged as two prominent approaches for the estimation of vegetation biochemistry. This study evaluates the performance of five machine learning regression algorithms (MLRAs) for the mapping of crop canopy chlorophyll at the Kellogg Biological Station (KBS) in Michigan, USA, across three scenarios: (1) application to Landsat 7, RapidEye, and PlanetScope satellite images; (2) application to UAV–satellite data fusion; and (3) integration with the PROSAIL radiative transfer model (hybrid methods PROSAIL + MLRAs). The results indicate that the majority of the five MLRAs utilized in UAV–satellite data fusion perform better than the five PROSAIL + MLRAs. The general trend suggests that the integration of satellite data with UAV-derived information, including the normalized difference red-edge index (NDRE), canopy height model, and leaf area index (LAI), significantly enhances the performance of MLRAs. The UAV–RapidEye dataset exhibits the highest coefficient of determination (R2) and the lowest root mean square errors (RMSE) when employing kernel ridge regression (KRR) and Gaussian process regression (GPR) (R2 = 0.89 and 0.89 and RMSE = 8.99 µg/cm2 and 9.65 µg/cm2, respectively). Similar performance is observed for the UAV–Landsat and UAV–PlanetScope datasets (R2 = 0.86 and 0.87 for KRR, respectively). For the hybrid models, the maximum performance is attained with the Landsat data using KRR and GPR (R2 = 0.77 and 0.51 and RMSE = 33.10 µg/cm2 and 42.91 µg/cm2, respectively), followed by R2 = 0.75 and RMSE = 39.78 µg/cm2 for the PlanetScope data upon integrating partial least squares regression (PLSR) into the hybrid model. Across all hybrid models, the RapidEye data yield the most stable performance, with the R2 ranging from 0.45 to 0.71 and RMSE ranging from 19.16 µg/cm2 to 33.07 µg/cm2. The study highlights the importance of synergizing UAV and satellite data, which enables the effective monitoring of canopy chlorophyll in small agricultural lands.
more » « less- Award ID(s):
- 2224712
- PAR ID:
- 10530536
- Publisher / Repository:
- MDPI
- Date Published:
- Journal Name:
- Remote Sensing
- Volume:
- 16
- Issue:
- 12
- ISSN:
- 2072-4292
- Page Range / eLocation ID:
- 2058
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
null (Ed.)Timely and accurate monitoring has the potential to streamline crop management, harvest planning, and processing in the growing table beet industry of New York state. We used unmanned aerial system (UAS) combined with a multispectral imager to monitor table beet (Beta vulgaris ssp. vulgaris) canopies in New York during the 2018 and 2019 growing seasons. We assessed the optimal pairing of a reflectance band or vegetation index with canopy area to predict table beet yield components of small sample plots using leave-one-out cross-validation. The most promising models were for table beet root count and mass using imagery taken during emergence and canopy closure, respectively. We created augmented plots, composed of random combinations of the study plots, to further exploit the importance of early canopy growth area. We achieved a R2 = 0.70 and root mean squared error (RMSE) of 84 roots (~24%) for root count, using 2018 emergence imagery. The same model resulted in a RMSE of 127 roots (~35%) when tested on the unseen 2019 data. Harvested root mass was best modeled with canopy closing imagery, with a R2 = 0.89 and RMSE = 6700 kg/ha using 2018 data. We applied the model to the 2019 full-field imagery and found an average yield of 41,000 kg/ha (~40,000 kg/ha average for upstate New York). This study demonstrates the potential for table beet yield models using a combination of radiometric and canopy structure data obtained at early growth stages. Additional imagery of these early growth stages is vital to develop a robust and generalized model of table beet root yield that can handle imagery captured at slightly different growth stages between seasons.more » « less
-
Abstract. As the changing climate expands the extent of arid andsemi-arid lands, the number of, severity of, and health effects associated with dust events are likely to increase. However, regulatory measurements capable of capturing dust (PM10, particulate matter smaller than10 µm in diameter) are sparse, sparser than measurements of PM2.5 (PM smaller than 2.5 µm in diameter). Although low-cost sensors couldsupplement regulatory monitors, as numerous studies have shown forPM2.5 concentrations, most of these sensors are not effective atmeasuring PM10 despite claims by sensor manufacturers. This studyfocuses on the Salt Lake Valley, adjacent to the Great Salt Lake, whichrecently reached historic lows exposing 1865 km2 of dry lake bed. Itevaluated the field performance of the Plantower PMS5003, a common low-costPM sensor, and the Alphasense OPC-N3, a promising candidate for low-costmeasurement of PM10, against a federal equivalent method (FEM, betaattenuation) and research measurements (GRIMM aerosol spectrometer model1.109) at three different locations. During a month-long field study thatincluded five dust events in the Salt Lake Valley with PM10 concentrations reaching 311 µg m−3, the OPC-N3 exhibited strong correlation with FEM PM10 measurements (R2 = 0.865, RMSE = 12.4 µg m−3) and GRIMM (R2 = 0.937, RMSE = 17.7 µg m−3). The PMS exhibited poor to moderate correlations(R2 < 0.49, RMSE = 33–45 µg m−3) withreference or research monitors and severely underestimated the PM10concentrations (slope < 0.099) for PM10. We also evaluated aPM-ratio-based correction method to improve the estimated PM10concentration from PMSs. After applying this method, PMS PM10concentrations correlated reasonably well with FEM measurements (R2 > 0.63) and GRIMM measurements (R2 > 0.76), andthe RMSE decreased to 15–25 µg m−3. Our results suggest that itmay be possible to obtain better resolved spatial estimates of PM10concentration using a combination of PMSs (often publicly availablein communities) and measurements of PM2.5 and PM10, such as thoseprovided by FEMs, research-grade instrumentation, or the OPC-N3.more » « less
-
more » « less
Grapevine rootstocks are gaining importance in viticulture as a strategy to combat abiotic challenges, as well as enhance scion physiology. Direct leaf-level physiological parameters like net assimilation rate, stomatal conductance to water vapor, quantum yield of PSII, and transpiration can illuminate the rootstock effect on scion physiology. However, these measures are time-consuming and limited to leaf-level analysis. This study used different rootstocks to investigate the potential application of aerial hyperspectral imagery in the estimation of canopy level measurements. A statistical framework was developed as an ensemble stacked regression (REGST) that aggregated five different individual machine learning algorithms: Least absolute shrinkage and selection operator (Lasso), Partial least squares regression (PLSR), Ridge regression (RR), Elastic net (ENET), and Principal component regression (PCR) to optimize high-throughput assessment of vine physiology. In addition, a Convolutional Neural Network (CNN) algorithm was integrated into an existing REGST, forming a hybrid CNN-REGST model with the aim of capturing patterns from the hyperspectral signal. Based on the findings, the performance of individual base models exhibited variable prediction accuracies. In most cases, Ridge Regression (RR) demonstrated the lowest test Root Mean Squared Error (RMSE). The ensemble stacked regression model (REGST) outperformed the individual machine learning algorithms with an increase in R2 by (0.03 to 0.1). The performances of CNN-REGST and REGST were similar in estimating the four different traits. Overall, these models were able to explain approximately 55–67% of the variation in the actual ground-truth data. This study suggests that hyperspectral features integrated with powerful AI approaches show great potential in tracing functional traits in grapevines.
-
The size and distribution of Phytoplankton populations are indicators of the ecological status of a water body. The chlorophyll-a (Chl-a) concentration is estimated as a proxy for the distribution of phytoplankton biomass. Remote sensing is the only practical method for the synoptic assessment of Chl-a at large spatial and temporal scales. Long-term records of ocean color data from the MODIS Aqua Sensor have proven inadequate to assess Chl-a due to the lack of a robust ocean color algorithm. Chl-a estimation in shallow and coastal water bodies has been a challenge and existing operational algorithms are only suitable for deeper water bodies. In this study, the Ocean Color 3M (OC3M) derived Chl-a concentrations were compared with observed data to assess the performance of the OC3M algorithm. Subsequently, a regression analysis between in situ Chl-a and remote sensing reflectance was performed to obtain a green-red band algorithm for coastal (case 2) water. The OC3M algorithm yielded an accurate estimate of Chl-a for deep ocean (case 1) water (RMSE = 0.007, r2 = 0.518, p < 0.001), but failed to perform well in the coastal (case 2) water of Chesapeake Bay (RMSE = 23.217, r2 = 0.009, p = 0.356). The algorithm developed in this study predicted Chl-a more accurately in Chesapeake Bay (RMSE = 4.924, r2 = 0.444, p < 0.001) than the OC3M algorithm. The study indicates a maximum band ratio formulation using green and red bands could improve the satellite estimation of Chl-a in coastal waters.more » « less
